Reviewers' comments:
Reviewer #1 (Remarks to the Author):
In the manuscript "Mechanical On
-
Chip Microwave Circulator”, the authors have reported the
experimental realization of frequency tunable microwave isolator/circulator. The experiment is
indeed very interesting and the data well understood and modeled in de
tail. It would generate a
lot of excitement, as an on
-
chip microwave isolator/circulator which could even work at the single
“photon” level. In my opinion, the manuscript is suitable for publication in Nature Communication,
after the authors have addressed
the following comments and questions:
1
.
Can the authors discuss more details about the advantages of such devices, especially for the
quantum information processing? The discussion in the manuscript is too simple. It will be really
helpful if the author
s can estimate some parameters for experiment, i.e. temperature,
cooperativity.
.
2. What’s the bandwidth of the isolator/circulator? Is there any more application of such small
bandwidth especially for the microwave system?
3. What’s the input power t
o drive the mechanics and get higher cooperativity? Is there any
nonlinear effect?
4. The figures are small and illegible to read. It can be improved to one row. And what’s the unit of
the Fig.2b & 3b?
Reviewer #2 (Remarks to the Author):
In the ma
nuscript “Mechanical On
-
Chip Microwave Circulator” the authors report on an experiment
based on the existing theoretical ideas, to implement a microwave isolator/circulator. Integrating
optomechanics with superconducting qubits, by using the same microwave
technology, is
interesting and could provide new possibilities. This work builds on the existing ideas and
experiments in cavity optomechanics (see ref. 26 and refs within). The authors demonstrate that
not only the operating frequency of the device can b
e tuned, but also the direction of the
isolation/circulation can be controlled. The experimental results are in a good agreement with
authors’ theoretical analysis. I think if the authors address the issues below, publishing the paper
in Nature communicati
on can benefit a wide range of audience.
1) In the second paragraph, where limitations of the existing circulators based on magneto
-
optic
effects have been discussed, no reference is presented.
2) I think it helps if the authors clarify what the opt
omechanical and electromechanical couplings
refer to in their setup.
3) While the physics of non
-
reciprocity is based on reservoir engineering and the presence of
mechanical loss, the intuitive picture behind non
-reciprocal mechanism is postponed after
the
presentation of formalism and the results. I think it’s helpful to present the physical picture first,
and then delve into a mathematical description.
4) It seems that in insets of Fig 3b the black arrow indicates the circulation direction. It would
be
useful to mention it in the caption.
Reviewer #3 (Remarks to the Author):
The primary innovation in the paper "Mechanical On-Chip Microwave Circulator" is the first
experimental electro-mechanical circulator that is potentially chip-compatible with other cryogenic
microwave components. This is in contrast to the various demonstrations of opto- (refs 24-26) and
electro-mechanical (refs 28 & 29, apparently 29 is now in print) _isolators_, which are two port
non-reciprocal devices, as opposed to three.
Not without good reason, non-reciprocal components are having a bit of a "moment" right now in
integrated micro- and quantum-devices in the optical and RF domains, due to the limitations of the
standard technologies based on permanent magnets, as mentioned in the manuscript. Thus, this
paper opens the door to a new technological approach (i.e. mechanical) to achieving non-
reciprocity, and has the potential to be of interest to a fairly wide audience in integrated micro-
and quantum-devices in the optical and RF domains. So I feel the topic matter and this paper in
general deserves to be published in a broad journal. The main limitation of the paper is that the
circulator's performance is extremely poor by almost every metric compared to already proven
Joseph junction-based RF circulators (refs. 4 & 5 are two, but not the only examples), and it's not
at all clear how the mechanical approach could ever out compete them. So, it deserves publication
for its basic science, even if I am very pessimistic about its technological impact.
Assuming the authors can adequately address my questions and comments further below, I would
generally support publication in Nature Communications.
1) A small thing, but I generally prefer to describe devices such as this one as "electro-mechanics"
not "opto-mechanics." Here the authors sort of use the terms interchangeably, which is not
uncommon in the field because they are so closely linked, but I wonder if it might prove confusing
to other readers.
2) Fig. 1c needs to be redrawn to make the electrical connectivity much clearer. I cannot tell at all
what is electrically connected to what in the nanowire device and the text didn't help me
understand what I couldn't see in the figure. In particular, I cannot tell at all where the Vdc is
applying its potential. I am very confused why "As expected, resonators 1 and 3 are tuned to
higher frequency due to an increased vacuum gap while resonator 2 is tuned to lower frequency."
As far as I can tell in Fig. 1c, resonators 2 & 3 look identical.
3) In figure 1a, it looks like there are two additional inductively coupled ports. What are these for?
4) Please state the overall footprint area of the device.
5) If the cooperativity employed in the main text figures was stated, I missed it. Similarly, Please
also state the pump power required to achieve this cooperativity in dBm. This required power
metric is helpful in comparing this technology to others, such as the JJ devices.
6) What is the relationship between the theory discussed at length in this paper to that described
in the theory papers referenced in refs 6, 22, & 23? What is the relationship between the theory in
this paper and the theory in the JJ-based components? For example, ref. 4 also involves
parametric frequency conversion between the three port modes and two internal modes to achieve
circulation. Is this approach and that one formally equivalent?
7) I want to see not just the forward and backward scattering matrix elements, but S11, S22, &
S33 as well. These input reflections are as important as the isolation for a usable device.
8) Similarly, I want to see an output PSD of all three ports while the component is circulating.
Again, as a practical matter, S
-parameters are not the only thing you care about in a c
irculator,
_especially_ an active one. This device employs six pump tones that are orders of magnitude
stronger than the signal tones and merely MHzs from the signal carriers. It is critically important to
know how bad all this RF leakage is.
9) I think
it would benefit readers if the authors provided more of an honest assessment of this
technology vis a vis other non
-
magnetic approaches. In particular, the JJ devices seems to out
perform this one in just about every way, and it's not at all clear if this
could ever change. For
example, this device provided ~10 dB relative isolation over merely ~300 Hz, while ref. 4 (the first
ever demonstration) had the same over 11 MHz. This component may be tuned by 30 MHz, while
that one could be tuned up to 400 MHz. T
his one had ~4 quanta of added noise, while that one
had 0.5 (quantum limited). This one requires six RF pumps, that one 3. Presumably the pump
power required here is again orders of magnitude greater than in ref. 4, because the nonlinearity
is so much wea
ker. If I had to guess, this mechanical technology might only be able to outperform
the JJ ones in 1 dB compression point, again because the nonlinearity is so much weaker.
As for the proposed application of superconducting qubit readout, the bandwidth o
f those signals
are measured in MHz, and carriers can easily vary by 100 MHz without a very well controlled
process. Even with its better parameters, the limitations of ref. 4 have yielded little actual
technological impact so far for this application eith
er. I want to reiterate that publishable work just
has to be innovative, rather than technologically competitive, but the authors can only help the
field by comparing it to other technologies and offering more guidance on where improvements will
be best ma
de.
We thank the referee for this positive comment about publication of our manuscript in Nature
communication.
1a
.
Can
the authors discuss more details about the advantages of such devices, especially for the
quantum
information processing? The discussion in the manuscript is too simple.
We have
added a more detailed description of the purpose of a microwave circulator for quantum
information
processing in the first paragraph of the introduction.
We have
also added the main advantages of a mechanical realization of a microwave circulator, in
parti
cular with respect to Josephson junction devices in the conclusion of the main text.
1.b
It will be really helpful if the authors can estimate some parameters for experiment, i.e. temperature,
cooperati
vity.
We thank
the referee for raising this point. In the main text we added the measured refrigerator
temperature
of 10 mK, the measured thermalization temperature of the mechanical modes of (18 / 25) mK
corresp
onding to a thermal phonon number of (85 / 90) for the mechanical modes 1 and 2. We also report
the sing
le cavity, single mechanical mode cooperativities for the pump powers used in the presented
isolator a
nd circulator experiments.
2a. What’s
the bandwidth of the isolator/circulator?
The
instantaneous bandwidth of the device is a combination of the effective mechanical damping rates
Γ